onboard

Developer onboarding is automated by running a guided interview that profiles your local environment, detects ML tooling, and writes a durable profile to ~/.understudy/profile.json. This configures subsequent skills to adapt to your machine and goals.

Local-first onboarding is a first-run setup process that downloads a small open-model, profiles your machine, and detects ML tooling locally. It creates a durable profile.json file that tailors future workflows without relying on external cloud services.

ML tooling detection and open-model setup require a local environment capable of downloading a small open model. The onboarding process verifies this readiness before proceeding with environment profiling and the guided interview.

To set up a durable local profile, run the onboarding flow after installing the Understudy plugin. It writes your machine profile, agent-card, and goals to ~/.understudy/profile.json, which future skills reference to adapt their workflows.

The best way to kick off first-run setup is to run the onboarding flow, which handles background model download, machine profiling, and tooling detection automatically. It hands off to the orchestrator to coordinate the broader workflow setup.

Onboarding writes to ~/.understudy/profile.json to store a durable record of your environment profile, ML tooling, and interview goals. This allows subsequent skills to reference and automatically adapt their behavior to your specific setup.